Monetary aggregates and economic activity: evidence from five industrial countries
BIS Economic Papers No 7
Since the early 1970s, a number of industrialised countries have pursued a strategy of money supply targeting as the primary focus of monetary policy. Essentially such a strategy has involved the setting of policy instruments to attain an intermediate target (the rate of growth of one or more of the monetary aggregates) with the ultimate aim of influencing the rate of growth of nominal income, particularly inflation, and perhaps affecting real economic variables also (Friedman 1975, OECD 1979, Saving 1967). As a result, the post-war Keynesian monetary approach based on interest rate stabilisation objectives, which allowed the money supply to be demand-determined, has generally been abandoned.
The major reasons behind the switch to monetary targeting are now well-known (Friedman 1982a, Dennis 1982). These included: i) the acceleration of inflation in the early 1970s and the difficulty of targeting on interest rates in such circumstances, ii) the considerable empirical support then prevailing for the key monetarist proposition of a stable demand for money (Laidler, 1981), iii) the expectation, with the advent of floating exchange rates, that small, open economies would be able to conduct an independent monetary policy and iv) growing dissatisfaction with the results of Keynesian demand-management (fine-tuning) policies.
More recently, however, some departures from this central theme have reflected new controversy over the appropriateness of strict monetary targeting in certain countries at the present time. For example, in September 1982, the Federal Reserve Board announced that it would for the time being pay less attention to M1 and would instead focus more on the broader monetary aggregates (M2 and M3). In 1982 also, Canada discontinued its M1 target while the United Kingdom, in dropping its exclusive attention on a Sterling M3 target, announced that it would in future monitor a range of aggregates as well as the exchange rate. The reasons behind these general modifications of policy are numerous. Policies may have been influenced by the deep recession in world demand and accompanying high unemployment since the second oil crisis of 1979 and by the threat to international financial stability arising from the debt servicing problems of many developing nations. More particularly, against a background of high and volatile interest rates, the demand for money in recent years has shown considerable instability in certain countries, not least the United States. The evolution of financial innovation, together with shifts in the demand for funds on precautionary grounds, have contributed to this instability. In consequence, monetary targeting procedures have become subjects of considerable debate, although the modifications that have occurred are no more than a partial retreat in a few countries. Thus, although the underlying foundations of a targeting strategy are slightly less secure, the advantages of such a strategy are still seen to be strong.
Within an intermediate targets strategy the optimal choice of the target monetary aggregate is typically made on the basis of three criteria (Andersen and Karnovsky 1977, Brittain 1981, Davis 1979, Friedman 1982b). Firstly, a monetary aggregate must be closely and stably related to the final goals of policy - usually summarised as the level or rate of growth of nominal income. This is the most important criterion and explains the considerable discussion in monetary analysis of the transmission mechanism between the money supply and nominal income (Laidler 1978). The second criterion concerns the strength and stability of the link between the spectrum of monetary instruments and the chosen monetary aggregate. (This also implies the need for the intermediate targets to be exogenous to non-policy developments or at least capable of having such factors reliably offset by actions of the monetary authorities.) Thirdly, for a monetary aggregate to be a feasible policy target, reliable information regarding its time path is required. Data on such an aggregate should, therefore, be accurate, frequently published and available with a short time-lag only.
The purpose of the present paper is limited. In view of the renewed debate on targeting and especially on the choice of target aggregates, it would seem useful to have a new look at the long-term relationships between nominal income and particular monetary aggregates across a selected group of countries. Hence the paper presents evidence on the first criterion above, that is the strength of the links between nominal income and a wide range of concepts of money and credit using data for the United States, the United Kingdom, Germany, France and Italy. The tests are extended by the addition of autonomous expenditure variables to the estimating equations. This is done to allow estimation of a formal reduced-form equation derived from a conventional model of an open economy, as the omission of such alternative arguments in the determination of nominal income would lead to biased coefficients on the monetary aggregates. However, it must be emphasised that the tests in this paper should not be viewed as a re-run of the familiar St. Louis equation. Thus there is little inference drawn in the paper on the relative role of monetary and autonomous spending variables in the determination of fluctuations in nominal income.
The major aims of the study are therefore twofold:
- to investigate the relationships between nominal income and alternative monetary aggregates in individual countries;
- to observe whether there are any cross-country patterns in the money supply nominal income relationships for certain common definitions of money such as narrow money, broad money or credit.
It is in the second respect, perhaps, that the paper is of main interest in the ongoing debate concerning the choice of the target aggregate. Few attempts have been made to provide a cross-country survey of a wide range of aggregates in this way. The inevitable cost is some absence of sophistication in the econometric evidence. Without doubt, a larger number of more complex specifications and tests could have been undertaken for the individual countries covered.
The main cross-country themes that are apparent from the regression results presented in this paper may he briefly summarised. Firstly, it is clear that the broadest concepts of liquidity or total credit, in those countries for which data are published, are closely related to nominal income. Secondly, and in contrast, credit from the banking system alone has a much less close relationship with nominal income for all countries except Germany and Italy. Thirdly, with respect to the conventional monetary aggregates, a broad measure (M3) is generally more closely related to nominal income than a narrow definition of money (M1), although the United Kingdom is an exception to this general result. Finally, the relationship between the monetary base and nominal income is well defined for the United States and United Kingdom, although slightly less so in the case of Italy. Turning to the velocity graphs, a common theme is that the velocity of MI has trended upwards over the observation period in all countries except Italy. In contrast, the velocity of broader concepts of money has generally fallen, with the exceptions of the United Kingdom, where the velocity of all aggregates has risen over the data period, and the United States, where the velocity trends for all broad concepts of money, and also for the Debt Proxy and Total Credit, have been effectively flat. Any international variations in results are likely to reflect many complex factors, but particularly institutional differences in payment systems and the extent of financial innovation, and also variations in the rates of financial saving and investment.
It must be emphasized that these results are based on behaviour over a long-run period. They are derived from a study of annual data over, approximately, twenty-year periods. The possibility that short run, reversible fluctuations in the nominal income to money relationships may have occurred which are not tracked in this study does not invalidate conclusions with respect to such secular behaviour. The results do, however, reflect any permanent movement in velocity relationships that may have occurred over the data period. However, any fundamental shifts in the relationships between monetary aggregates and nominal income that, for regulatory or structural reasons, may be evolving at present are not necessarily picked up in a historical study of this type.
In addition, the results in this paper stand on their own as evidence on the first, and most important, criterion governing the choice of targets. However, any unambiguous, broadly-based conclusions would also require evidence on the relative controllability of the aggregates in question. Therefore, this paper's concentration on the strength of the links between money and nominal income should not be taken to mean that the other criteria - especially controllability - are unimportant.